The identification of a reliable MRI marker and the structural background of clinical and cognitive symptoms in MS are still hot topics. While lesions in the white matter are remarkable MRI markers in MS and counted as one of the diagnostic cornerstones of the disease, the lesion load only modestly correlates with clinical and cognitive decline. This phenomenon is known as the clinico-radiological paradox. Although the Gd-enhancing lesions are slightly more specific MRI markers during the acute phase of the disease, they can rarely be seen on the MRI. It can happen, even in cases of relapse, that no Gd-enhancement appears. Since the Gdenhancement can only be monitored for 2 or 3 weeks, detection is barely possible during a routine MRI. GM atrophy (cortical with or without subcortical structures) can be detected by several approaches. Diffusion is the movement in the material, which is not accompanied by huge molar motions. It can be non-invasively portray by DTI. Cellular elements (membranes) inhibit molecular diffusion, thus the architecture of the tissues can be detected by the diffusion profile of water. Tract-Based Spatial Statistics (TBSS) and the manual labeling of the regions-of-interest (ROIs) are useful approaches for localised statistical testing of FA (and other diffusion-related) data. In our first study we aimed to determine the relationship between cortical atrophy and WM pathology in MS. Within this, we studied whether the pathology of the focal lesions or the diffuse NAWM has a more significant role in the evolution of GM atrophy. Based on Jenha’s study, we were also interested in the connection between the desintegration of the periventricular WM and the conception of GM atrophy. Our aim was to test two hypotheses. According to the first, GM atrophy is defined by demyelination-like diffusion features that suggest there is a common root in the development of demyelination in WM and GM atrophy (maybe a common pathological process mediated by the CSF). The second hypothesis proposed that GM atrophy is more connected to axon loss-like diffusion pattern, which points to a role of remote axonal transection in GM loss. In our second study, we meant to define the pattern of the MRI parameters best predicting clinical and cognitive disability in patients with MS. Since MRI parameters are strongly related, the conventional linear regression analysis is not an appropriate choice for the statistics. In our studies, we used the model-free partial least square 6 (PLS) approach, which can detect the design of those parameters that best predicts the questionable parameters, besides dealing with the problem of collinearity.